View
3
Download
0
Category
Preview:
Citation preview
1
Head-marking inflection and the architecture of grammatical theory:
Evidence from reduplication and compounding in Hiaki (Yaqui)
Jason D. Haugen Heidi Harley
Oberlin College & University of Arizona
jhaugen@oberlin.edu hharley@u.arizona.edu
Abstract:
Within generative grammar, noun incorporation and other compounding processes have
traditionally been the focus of morpho-syntacticians, while reduplication has been investigated
primarily by morpho-phonologists. The interaction of these two phenomena in a single language
has significant implications that go beyond the narrow concerns of these two sub-domains,
bearing much more broadly on the architecture of grammatical theory. This paper investigates
the interactions of reduplication and compounding within one language, Hiaki (Yaqui).
Reduplication for aspectual inflection in Hiaki occurs inside of compounds and other derived
words, marking the head of the word. We demonstrate the major architectural issues resting on
the analysis of these phenomena by examining how different theoretical perspectives can (or
cannot) accommodate the Hiaki data. Architectures which separate inflection and derivation into
separate grammatical components (Weak Lexicalist theories) cannot account for the data, while
theories maintaining the two processes in the same component fare better but tend to require
some supplemental architectural accommodations. Strong Lexicalist theories and Paradigm
Function Morphology can account for head-marking reduplication in Hiaki, but other data from
the language problematize major premises of these theories. The syntacticocentric approach we
advocate also requires supplementation (Local Dislocation) to correctly position inflectional
reduplicants inside compound verb forms, but this is a well-established and independently
motivated mechanism within the Distributed Morphology framework.
Keywords:
Inflection, head-marking, reduplication, noun incorporation, compounding, Hiaki (Yaqui)
2
Head-marking inflection and the architecture of grammatical theory:
Evidence from reduplication and compounding in Hiaki (Yaqui)
Jason D. Haugen Heidi Harley
Oberlin College & University of Arizona
jhaugen@oberlin.edu hharley@u.arizona.edu
1. Introduction
Two areas that have received an extraordinary amount of attention in contemporary theory-
building have been reduplication, at the interface of morphology with phonology, and noun
incorporation (NI), at the interface of morphology with syntax. Given the wide cross-linguistic
distribution of these two phenomena in addition to their great theoretical importance, it is
relatively surprising that the interaction of the two has received extremely little (and in fact,
almost no) attention in the literature. The purpose of this paper is to provide extensive empirical
documentation of how reduplication interacts with noun incorporation and other types of verb
compounding in one language, Hiaki (Yaqui). We will show that V(erb)-V(erb) compounds
allow reduplication to apply to either (or both) of the verbal elements, whereas noun
incorporation constructions, which are in essence N-V compounds, only allow for reduplication
as a prefix to the verbal element. Inflectional reduplication thus appears inside the compound in
noun incorporation contexts.
We then go on to address the theoretical ramifications that these data have for a variety of
different frameworks in theoretical morphology. The central issue raised by the interaction of
reduplication with compounding in Hiaki is word-internal head-marking inflection. As we will
show, the Hiaki reduplication + compounding data, and word-internal head-marking more
generally, raise interesting problems for competing theories. We will argue that those theories
whose architectures contain inflectional and derivational processes in a single grammatical
3
component (e.g. Strong Lexicalism or syntacticocentric theories like Distributed Morphology)
fare far better in accounting for these data than those which would separate the two into separate
components (e.g. Weak Lexicalist theories). We will also argue that a recent theory designed
explicitly to account for head-marking inflection, Paradigm Function Morphology (Stump 2001),
can straightforwardly account for the basic Hiaki reduplication facts but runs into empirical
problems arising from the interaction of reduplication and compounding with yet other aspects of
Hiaki grammar.
Before proceeding to our main discussion, however, we would first like to say how
thrilled we are to include our paper in this volume in honor of the one and only Jane Hill. She
has been an exemplary scholar and wonderful personal friend to both of the present authors (not
to mention her service as co-advisor on Haugen’s 2004 University of Arizona dissertation). No
single paper could hope to encompass all of the many voices of Jane Hill—indeed, it is our belief
that no single person could match the range of topics that she has covered with the depth and
insight that she has contributed to so many branches of linguistics, anthropology, and fields
beyond. The particular “voice” of Jane Hill with which we hope to connect our present paper is
the one that applies linguistic data collected from the documentation of threatened and
endangered languages to empirical considerations crucial to linguistic theorizing—i.e. what Ken
Hale (2000) has called the “confirmatory function” of linguistic diversity. Along these lines
Jane’s own work on prosodic morphology and word derivation in Tohono O’odham (Hill and
Zepeda 1992) and the morphology and semantics of reduplication in Tohono O’odham (Hill and
Zepeda 1998), as well as her work on the theoretical ramifications of the complex verb
morphosyntax of Cupeño (Hill 2003), spring immediately to our minds. We hope that our
4
remarks on the theoretical ramifications of reduplication and compounding in Hiaki will be taken
in a similar spirit.
We’d also like to thank Jane for all of the work that she has done to foster at the
University of Arizona a unique and exciting research community of scholars working on all
aspects of Uto-Aztecan linguistics, through which we have both profited greatly—intellectually,
personally, and otherwise. This one’s for you, Jane!
This paper is organized as follows. Section 2 outlines how the two different morphological
processes of reduplication and noun incorporation (N-V compounding) each work in Hiaki.
Section 3 then discusses the interaction of Hiaki reduplication and compounding processes in
several different contexts: NI constructions (§3); in hybrid verb constructions (§3.1); with verbal
suffixes (§3.2); and in “pseudo”-compounds (§3.3). Section 4 then focuses on the implications
of the word-internal reduplication process that occurs specifically in NI constructions (and
pseudo-compounds), in standard theoretical architectures: Weak and Strong Lexicalist
approaches and the syntacticocentric approach to incorporation proposed by Baker (1988). We
conclude that the order of the inflectional reduplicative morpheme in Hiaki, which appears word-
internally marking the verbal head of the compound, cannot be accounted for in the Weak
Lexicalist theory or in Baker’s theory without additional architectural accommodations.
Section 5 then discusses some architectural accommodations that have been proposed
which can account for these data, in both Lexicalist and syntacticocentric frameworks. We argue
that while the Hiaki reduplication data specifically are consistent with the predictions of the
Paradigm Function Morphology proposed by Stump (2001), its interaction with other cases of
inflection violate a crucial principle of PFM—namely, inflectional object clitics exhibit edge-
marking behavior in the same forms in which reduplication exhibits head-marking, in violation
5
of PFM's Paradigm Uniformity Generalization. We close by discussing how the Hiaki data can
be accounted for in a syntacticocentric framework by amending Baker’s theory with the notion
of Local Dislocation, within the theory of Distributed Morphology. Section 6 concludes.
1.1 Background: Reduplication and compounding in the literature
Reduplication interacting with the general phenomenon of compounding has received some
recent attention, and we identify two general types of previous research in this area. The first, of
particular importance in the present context, is reduplication plus compounding. Recent work in
this area includes discussion of reduplication in nominal compounds in Pima (Munro and Riggle
2004), a Uto-Aztecan language of the Tepiman sub-group (see footnote 12 below), as well as
Haugen (2010), which provides a more general discussion of reduplication in compounding
contexts cross-linguistically.
A second strand of research has examined the process of reduplication itself as a kind of
compounding (e.g. Zoll 2002; Downing 2003; Inkelas and Zoll 2005). For example, Downing
(2003) proposes that reduplicated verb stems in Bantu are composed of a compound stem formed
by a reduplicated stem combining with the base stem, as in (1):
(1) Compound Structure for Reduplicated Bantu Verb Stems (Downing 2003: 7 [7b])
Verb Word 3 INFL MacroStem 3
(OM) [Compound Stem]Stem1 3
[RED Stem]Stem2 [Base Stem]Stem3
Inkelas and Zoll (2005) provide a similar representation in Morphological Doubling Theory
(MDT), but they propose instead that a reduplicant and its “base” are simply potentially
6
heterogeneous daughters of a morphological mother node. The representation that they give for
reduplication structures is shown in (2):
(2) Schematic for Reduplication in Morphological Doubling Theory (I&Z 2005: 19 [27])
[zzz]
2 Co-phonology Z
[xxx] [yyy]
Co-phonology X | | Co-phonology Y
/Stemi/ /Stemi/
Our focus in this paper will involve only the first kind of investigation, as we will be
examining the interactive processes of inflectional reduplication occurring in the contexts of
noun incorporation and other compound verbs in Hiaki.
2. Reduplication and Noun Incorporation in Hiaki
Hiaki (Yaqui) is a Southern Uto-Aztecan language of the Taracahitic sub-group, and is
indigenous to northwestern Mexico (Sonora) but also spoken in southern Arizona, USA. Hiaki
allows for the productive compounding of verbal roots to indicate various semantic notions.
These compounds do not generally allow for any other constituent or affix to intervene between
their components, with one exception: reduplication.
2.1 Reduplication in Hiaki
Reduplication in Hiaki is a productive inflectional process typically marking habitual or
progressive aspect (also, in some cases, emphasis) on verbs, and it is generally prefixal. There
are several forms of reduplication which can be used for any of the above semantic functions: a
light syllable reduplicant; a heavy syllable reduplicant that triggers gemination of the onset of the
base into the coda of the reduplicant; a disyllabic reduplicant; and a word-internal pattern of
morphological gemination. In general, neither the form nor the meaning of a reduplication type is
7
fully predictable based on the phonological makeup of the stem to which it attached. We will not
concern ourselves here with this rampant reduplicative allomorphy; for further discussion see
Harley and Amarillas (2003), Haugen (2003), and Harley and Leyva (2009). Examples
illustrating the three most common meanings associated with one of the reduplication types, a
light syllable, are presented in (3) (the reduplicant will appear in bold in all examples
henceforth):1
(3) a. Habitual (HAB)
Itepo hunum ke-ke’ewe
1.PL there RED-gather.firewood
‘We gather firewood there.’
b. Progressive/continuative (‘in progress’), (PROG)
Uu hamut totoi kava-m bwa-bwata
The woman chicken egg-PL RED-stir
‘The woman is mixing the eggs.’
c. Emphatic (EMPH) (often in Imperative (IMP) examples)
Kat=ee uka soto'i-ta hunum ma-mana
NEG.IMP=2.SG the.ACC pot-ACC there RED-put
‘Don’t put that pot there.’
(Harley & Leyva 2009: 253 [13])
2.2 Noun incorporation in Hiaki
Noun incorporation involves the compounding of a noun with a verb, where in the usual case the
noun satisfies the verb’s internal argument selectional properties. A typical Hiaki example is
given in (4) below:
(4) Peo maso-peu-te-k
Peo deer-butcher-INTR-PERF
‘Peo deer-butchered’ (Jelinek 1998: 213 [48])
1 The abbreviations for our glosses are as follows: 1 = 1
st person; 2 = 2
nd person; 3 = 3
rd person; ACC = accusative; AF
= unsepecified affix; AG.NOM = agentive nominalizer; APPL = applicative; CAUS = causative; CONT = continuative;
DESID = desiderative; DET = determiner; DIR = directive; EMPH = emphatic; FUT = future; HAB = habitual; INCEP =
inceptive; INST = instrumental; INTR = intransitive; NEG.IMP = negative imperative; OBJ = object; OBL = oblique; PERF
= perfective; PASS = passive; PAST = past tense; PL = plural; PPL = past participle; PRET.AUG = preterite augmentative;
PROSP = prospective; RED = reduplication; REV = reverential; SG = singular; SUBJ = subject; TO = directional
postposition; TRAN = transitive; WH = question word.
8
Hiaki is an SOV language, so objects normally precede their verbs in any case. There are,
however, clear diagnostics that distinguish incorporated objects from non-incorporated O-V
juxtapositions. First, an incorporated nominal in Hiaki does not indicate number or case, which
contrasts with object nouns in verb-external object noun phrases, which must do so—compare
the inflection on the nominal stem maaso ‘deer’ in (5) with the absence of inflection on its
incorporated counterpart in (4).2
(5) a. Peo maso-ta peu-ta-k
Peo deer-ACC.SG butcher-TRAN-PERF
‘Peo butchered a deer’
b. Peo maso-m peu-ta-k
Peo deer-PL butcher-TRAN-PERF
‘Peo butchered some deer’
Second, no constituent may intervene between a noun and the verb in an NI construction,
as seen by comparing (6)a' with its overtly transitive counterpart in (6)b'; the adverbial aman in
the latter can intervene between the verb and its inflected object, but not between the verb and
the incorporated, uninflected nominal in the former.
(6) a. kuta-siu-te a'. *kuta-aman-siu-te
stick-tear-INTR stick-there-tear-INTR
‘wood-split’ ‘wood-split over there’
b. kuta-m siu-ta b'. kuta-m aman siu-ta
stick-PL tear-TRAN stick-PL there tear-TRAN
‘split wood.' ‘split wood over there’
The above examples show that the integrity of the noun-verb complex in Hiaki NI cannot be
disrupted by another syntactic constituent, such as a locational adverb.
Third, in some cases, these Hiaki NI constructions involve nominal stems rather than
fully free nominals. In Hiaki, some stems have special forms which are used when the stem is
2 The underlying vowel in maaso ‘deer’ is long and would surface as such in, e.g. the nominative case, which is
unmarked; long vowels in Hiaki stems regularly shorten under suffixation.
9
subject to derivational affixation; these are distinct from the free stems which are typically used
in inflectional affixation (see Tubino Blanco and Harley 2010 for a full discussion). The NI
examples in (7) show that these special derivational stems are used in NI constructions, and are
hence easily distinguished from the corresponding verb phrases with independent NP objects,
which would use the free stem form for the object. The incorporated forms in (7) are derived
using the bound stems of the nominals whose free stems are chichi ‘saliva’ and hipetam ‘bed’,
respectively:
(7) a. chit-wat-te ‘spitting’
saliva-throw-INTR derived from chichi ‘saliva’ n. + watta ‘throw (tr.)’
b. hipe-teka ‘make bed’
bed-lay.across.TRAN derived from hipetam ‘bed’ + teéka ‘lay sthg. across’
Fourth, word-internal phonological processes apply inside these NI constructions in
Hiaki, showing that they form a single phonological word. For example, they undergo the word-
internal phonological rule of /s/ → [h] in word-medial syllable final position. (8) illustrates this
process in a normal affixation context, using the verb stem bwasa/bwase ‘cook (tr./intr.)’; when
suffixed with the future suffix –ne in (8)b, the stem-final /s/ becomes [h]:
(8) a. Haisa intok bwa-bwasa’a-wa?
how and RED-cook-PASS
‘And how are they cooked?’ (Conversation 8 [15])
b. abwe, oowa-m ae-t mo-monto-wa hunak veha bwah-ne
well coal-PL it-on RED-pile-PASS then then cook-FUT
‘Well, coals are placed on it and then it will cook’ (Conversation 8 [16])
This rule also applies in NI constructions, illustrated in (9). The noun stem lioh ‘God’ in (9)a is
derived from Hiaki Lios, which in turn was borrowed from the Spanish dios ‘God’. It appears as
Lios in non-compounded contexts (9)b:
(9) a. lioh-bwania
God-promise
10
‘giving thanks’
b. Lios enchi ania.
God you.ACC help
‘(May) our creator help you’ (a traditional greeting)
Fifth, these N-V compounds are conducive to idiomatic interpretation, and are often used
for common, culturally-relevant activities, as is characteristic of compounding NI constructions
more generally (Escalante 1990:105, Dedrick and Casad 1999:161) (cf. Mithun 1984):
(10) a. tekil-maka ‘commissioning, making responsible’ ( < lit. ‘work-giving)
b. kuchu-sua ‘fishing’ ( < lit. ‘fish-kill.PL.OBJ.’ )
Hiaki noun incorporation, then, appears to produce normal N-V compounds, exhibiting
all the appropriate word-like properties expected from the result of a derivational process.
Mithun (1984) differentiates two sub-types of what she calls Type 1 NI: true compounding NI,
which creates a single verbal word, and composition by juxtaposition, which exhibits a much
looser morpho-phonological connection between the two elements of the compound. Hiaki NI
constructions must be regarded as of the former type, since NI compounds meet all the
identificational criteria laid out by Mithun.
2.3 Hiaki NI as detransitivizing?
Rosen (1989) proposes a major distinction among NI types along the lines of transitivity—i.e.
intransitive versus transitive NI constructions (compound vs. classifier NI, in her terminology).
Hiaki NI has generally been taken to be intransitivizing (Escalante 1990, Jelinek 1998), as
illustrated in the following contrasting examples presented by Jelinek (1998: 213 [48]), repeated
from (4) and (5) above:
(11) a. aapo maso-ta peu-ta-k
3sg deer-ACC butcher-TRAN-PERF
‘He butchered a deer.’
11
b. aapo maso-peu-te-n
3sg deer-butcher-INTR-PAST
‘He was deer butchering.’
c. *aapo bwe’uu-k maso-peu-te-n
3sg big-ACC deer-butcher-INTR-PAST
[*‘He was [big deer]-butchering.’] or [*‘He was deer-butchering a big one.’]
(11)a presents a transitive sentence with an accusative case-marked direct object nominal
(masota ‘deer-ACC’) and a verb marked with the transitive suffix –ta. (11)b provides the
corresponding NI example, where the nominal is compounded with the verb (and is no longer
marked with the accusative marker), and the verb takes the intransitive suffix –te. That such NI
verbs are truly intransitive is further demonstrated in (11)c, which illustrates the
ungrammaticality of external modifiers (adjectives, numerals, determiners, etc.) with
incorporated nominals.
However, in contrast to Jelinek’s data, in some cases it appears to be possible to have
“stranded” or (“null-head”) modifiers with incorporated nominals in Hiaki. Molina et al. (1999)
list pan hoa as an intransitive verb meaning ‘to make bread’ (< pan ‘bread’ + hooa ‘make’). In
(12) we present new empirical evidence that this is actually a transitive NI construction:
(12) a. Irene panim am-hoo-ria
Irene bread-PL 3.PL-make-APPL
‘Irene is making bread for them.’
b. Irene am=pan-hoo-ria
Irene 3.PL=bread-make-APPL
‘Irene is making bread for them.’
c. *Irene pan am=hoo-ria
Irene bread 3.PL=make-APPL
d. Irene oficiom sii kiam pan-hoo-ria
Irene oficio very delicious-PL bread-make-APPL
‘Irene is making very delicious bread for the ceremonial officials’
12
(12)b and c show that the nominal root pan ‘bread’ must be incorporated onto the verb, given
both the inability of the third person plural agreement clitic am= (which specifies the benefactee
argument introduced by the applicative suffix –ria) to intervene between it and the verb hooa,
and the absence of inflection on pan. However, in (12)d, the intensifier and adjective modifying
pan show that the incorporated nominal can still be externally modified in some instances, which
is a sign of the transitivity of this construction in Rosen’s typology.
In theories which posit that an incorporated nominal forms a constituent with its external
modifiers, either through movement (e.g. Baker 1988) or co-analysis (Sadock 1991), such
modifiers are considered to be “stranded”. Indeed, such “stranding” of modifiers is presented as
a motivation for these theories. Theories like that of Rosen 1989, on the other hand, regard
incorporation as a (non-syntactic) morphological compounding process, and regard the modifiers
in the external NP to be “null-head” phrases. We’ll return to the differing predictions of these
two contrasting approaches below. In any case, the evidence suggests that Hiaki NI is at least
sometimes transitive NI, in Rosen’s terms, though often intransitive NI in other cases.
We now turn to illustrate the interaction of NI and reduplication in the language: What
happens when a compound V is inflected for habitual aspect by means of reduplication?
3. Reduplication with Noun Incorporation in Hiaki
Above we saw that nothing could intervene between the nominal and the verb stem in Hiaki NI.
The only exception to this generalization is the case of reduplication, when reduplication is used
as the means to express a verbal inflection, as illustrated in (13):
(13) Peo maso-peu-peute
Peo deer-RED-butcher
‘Peo is always butchering deer’
13
Here, inflection appears to disrupt the lexical integrity of the N-V compound. In NI in Hiaki,
reduplication cannot target the left edge of the compound:
(14) *Peo *ma-maso.peute / *Peo *maso-maso.peute
Peo RED-deer.butcher
‘Peo is always butchering deer’
A number of cases of reduplication in Hiaki noun incorporation (NI) constructions are listed in
(13). These examples all involve a nominal stem being compounded with an identifiable verbal
stem, i.e. a verbal element that can also appear as a free verbal stem without an incorporated
nominal.
(15) Verb Meaning Reduplicated Form
a. chit-wat-te ‘spitting’ chit-wat-watte
saliva-throw-INTR
b. hiavih-muke ‘gasping, short of breath, suffocating’ hiavih-mu-muke
breath-die (sg. subj.)
c. hipe-teka ‘make bed’ hippe-te-teka
bed-lay.across(t.v.)
d. kuchu-sua ‘fishing’ kuchu-su-sua
fish-kill (PL.OBJ.)
e. kova-hamti ‘deep in thought, concentrating, thinking’ kova-ham-hamti
head-broken
f. kupi-tomte ‘lose sight temporarily’ kupi-tom-tomte
eye-blossom
g. lioh-bwania ‘giving thanks’ lioh-bwa-bwania
God-promise
h. lio-noka ‘praying’ lio-no-noka
God-talk
i. Mao-noka ‘speaking in the Mayo language’ Mao-no-noka
Mayo-speak
j. masa-vaite ‘flapping wings’ masa-vai-vaite
14
wing flap
In particular, note that this process applies to cases in which the N-V compound involves a
bound nominal stem, as in (15)a and c; that no inflection appears on the nominal half in the
reduplicated form; and that word-internal phonological processes still apply to the entire N-V
compound with intervening reduplication, as in (15)g.
The resulting complex reduplicated NI form continues to behave as a V with respect to
other inflectional processes of the language. To illustrate, consider the following example, which
shows that the preverbal object clitics attach to the left edge of the compounded N-V form at the
same time that reduplication is marked internal to the compound (16)a; reduplicating at the left
edge of the compound N-V form is ungrammatical (16)b. Similarly, attaching the object clitic to
the reduplicated V, between the incorporated N and its sister V, is ungrammatical (16)c:
(16) a. Irene am=pan-ho-hoo-ria
Irene 3.PL-bread-RED-make-APPL
‘Irene is always making bread for them.’
b. *Irene am=pa(n)-pan-hoo-ria
Irene 3.PL- RED-bread-make-APPL
c. *Irene pan-am=ho-hoo-ria.
Irene bread-3.PL=RED-make-APPL
Reduplication in combination with NI in Hiaki, then, applies to the head V within the
otherwise apparently completely atomic N-V compound.
We next turn to illustrate extensively the interaction of reduplication with other verbal
compounding and derivation processes in Hiaki, first in derivations with “verb-affix hybrids”
(§3.1); then in other cases of verb-derivation, such as occurs with the affixation of derivational
suffixes on verbs (§3.2); and finally in words with unidentified word-internal structure (‘pseudo-
compounds’; §3.3).
15
3.1 Reduplication and Verb-Affix Hybrids
In addition to the reduplication of N-V compounds that occurs in NI constructions, Hiaki has a
small, closed class of free verb roots that can also be used as verbal suffixes. Under one possible
analysis, such constructions could be considered to be verb-verb compounds. We refer to this
closed class of verbs as “verb-affix hybrids”.3 Examples showing both free and compounded
uses of maachi ‘appear’ are presented in (17), and a pair illustrating free and bound uses of siime
‘go’ are presented in (18):
(17) a. Hai=sa maachi huu’u ‘em sa’awa
how=WH appear that your sore
‘How does your sore seem?’ Or ‘how is your sore?’
(Dedrick and Casad 1999: 67 [39])
b. Kaita-e mo’iti-machi
nothing-INST plow-appear
‘There is nothing with which to plow’ (Dedrick and Casad 1999: 67 [40])
(18) a. Yoko=ne potam-meu sim-ne
tomorrow=I Potam-TO go-FUT
‘I’m going to Potam tomorrow.’ (Dedrick and Casad 1999: 293 [1])
b. Inepo ili hu’unee-sime
I little know-go
‘I’m beginning to understand a little bit.’ (Dedrick and Casad 1999: 294 [7])
These V-V compounds exhibit properties similar to those illustrated above for N-V
compounds with respect to the use of bound stem forms for the left-hand member, word-internal
phonological alterations, and absence of internal inflectional material; they show no evidence of
being derived by a distinct word-formation process from the NI cases.
Reduplication with such verb-affix hybrids functions in the same way as with the NI
compounds: it appears word-internally, reduplicating the rightmost member of the V-V
compound. We illustrate the word-internal, head-reduplication pattern with our example hybrid
3 These behave identically to verbal affixes (suffixes) with respect to binding and the assignment of case in
subordinate clauses (see Tubino et al. 2009 for discussion).
16
verbs maachi ‘seem, appear’ and siime ‘go' in (19) below; Escalante 1990:78 also emphasizes
similar cases:
(19) a. Vempo si kuhti-ma-machi
3pl EMPH angry-RED-seem
‘They really seem like hateful people’.
b. Hita=sa empo hoo-si-sime
what=WH you do-RED-go
‘What are you going around doing?’
Unlike the case for reduplication with NI, however, reduplication in these V-V compounds can
target either member of the compound, so long as the semantics of reduplication can be applied
to either member. An example of reduplication applying to the first verbal element of a verb-
affix compound is given in (20), and an example with both members reduplicating is given in
(21).
(20) Vempo si kuh-kuhti-machi
3pl EMPH RED-angry-seem
‘They seem like really hateful people.’
(21) Vempo si kuh-kuhti-ma-machi
3pl EMPH RED-angry-RED-seem
‘They really seem like really hateful people.’
3.2 Reduplication and Verbal Affixes
We see the same word-internal, head-reduplication pattern on certain affixes, where
reduplication occurs between a verb stem and a derivational suffix. In these cases, it is the suffix
which receives the reduplication, despite the suffix being a bound element which otherwise does
not occur as an independent verb in the language. Some examples are presented in (22):
(22) a. Directive (-sae) (Escalante 1990 : 78 [41])
inepo a=nok-sae → inepo a=nok-sas-sae
1sg 3s=talk-DIR 1sg talk-RED-DIR
‘I am telling him to speak up’ ‘I tell him to speak up’
17
b. Desiderative (-’ii’aa) (Escalante 1990 : 78 [42])
inepo a=nok-’ii’aa → inepo a=nok-’ii-’ii’aa
1sg 3s=talk-DESID 1sg 3sg=talk-RED-DESID
‘I want him to talk’ ‘I would like him to talk (more)’
c. Inceptive (-taite) (Escalante 1990 : 79 [43])
aapo nok-taite → aapo nok-tai-taite
3sg talk-INCEP 3sg talk-RED-INCEP
‘She is starting to talk’ ‘S/he repeatedly starts to talk (hesitates)’
d. Prospective (-vae) (Escalante 1990: 79 [44])
aapo nok-vae → aapo nok-vav-vae
3sg talk-PROSP 3sg talk-RED-PROSP
‘He feels like talking’ ‘From time to time he gets the urge to talk’
‘S/he gets the desire to talk a lot’
These verbal affixes are always bound, and cannot appear as independent verbs at all, at
least in the synchronic grammar. That such affixes may not occur independently as the main verb
in a sentence is illustrated for a subset of them in (23) below:
(23) a. *Inepo apo’ik ii’aa
1.SG 3.SG.ACC want.
Intended reading: ‘I want it.’
b. *Aapo taite-k.
3.SG begin-PERF
Intended reading: ‘S/he began.’
c. *Aapo a=vae.
3.SG 3.SG =PROSP
Intended reading: ‘He'll do it/He feels like doing it.’
When reduplication applies to such suffixes, as above, it takes scope over the whole complex
verb form. In (24)a, for example, the meaning is one of habitual wanting-him-to-talk, not one
about habitual talking. If the speaker wishes to indicate habitual semantics for just the leftmost
member of the compound, reduplication applies to that leftmost member and takes scope only
over the lower verb; in (24)b, the speaker wants him to habitually talk, but doesn’t habitually
18
want anything. If the situation warrants it, habitual semantics and reduplication can apply to both
the stem and the suffix of the complex verb form (24)c:
(24) a. Inepo aa=nok-ii-’ii’aa ne vetchi’ivo
I him=speak-RED-want me for
‘I always want him to speak for me.’
b. Inepo aa=no-nok-ii’aa
I him=RED-speak-want
‘I want him to be the speaker/the one who habitually speaks.’
[e.g. at a council meeting]
c. Inepo aa=no-nok-ii-’ii’aa
I him=RED-speak-RED-want
‘I always want him to be the speaker.’
It is worth noting, however, that not all derivational verbal affixes can reduplicate. There
are some affixes that only allow reduplication to apply to the verbal stem. These include the
applicative -ria (25)a, the causative -tua (25)b, and a morpheme (-te) that derives verbs of
creation from nominal stems (25)c:
(25) a. Applicative (-ria)
lu-luuta-ria *luuta-ri-ria
RED-finish-APPL
‘(habitually) use up on someone’
b. Causative (-tua)
mahhai-tua *mahai-tu-tua
“RED”(μ-INFIX)-afraid-CAUS
‘really scare someone’
c. MAKE (-te)
hi-hipe-te *hipe-te-te
RED-mat-MAKE
‘(habitually) make mats’
We return to these in section 5.2 below in which we discuss available theoretical treatments.
In sum, many complex verbs in Hiaki can exhibit word-internal reduplication, as with the
N-V compounds illustrated above. However, unlike the complex verbs created by noun
19
incorporation, reduplication of verb stems with certain verbal suffixes allow three possibilities
for reduplication, with the reduplication taking scope over the morphological target: i.e. the verb
stem, the suffix, or both.
3.3 Word-Internal Reduplication in Pseudo-Compounds
There are also a variety of forms which display the pattern of internal reduplication, but which
do not involve otherwise free nominal or verbal stems. In fact, the synchronic internal make-up
of these forms is not known to us, since we only find the morphological elements in these
“pseudo-compound” contexts. In other words, these constructions are composed of cran-
morphs. A list of such pseudo-compounds is given in (26), along with the reduplicated form,
and in some cases, a possible etymological connection:4
(26) Verb Meaning Reduplicated Form Possible etymology
a. bwah-suma ‘braid’ bwah-su-suma < suma ‘tie’?
b. bwal-wotte ‘feel weak’ bwal-wot-wotte < bwala ‘sheep’?
c. bwal-wotta ‘make to feel weak’ bwal-wot-wotta < bwala ‘sheep’?
d. chiki-pona ‘tickling someone’ chiki-po-pona < poona ‘strike, knock’
e. le-siki-le ‘itching, tickles’ ele-si-sikile < siki ‘red’ ?
f. ha’a-chih-te ‘sneezing’ ha’a-chih-chihte < chitei ‘mash’ ?
g. haawahsa’a-te ‘steaming’ haa-wa-wahsa’ate < haawa ‘steam’
h. hun-hiawa ‘make fun of, tease’ hun-hi-hiawa
i. hu’u-nakte ‘created deliberately’ hu’u-na-nakte
4 While the majority of these forms may only reduplicate on the rightmost member, with left-edge reduplication
ruled flatly ungrammatical, our consultant felt that a couple of these forms could also accept reduplication at the
edge when prompted, e.g. (26)h. However, there was no semantic differentiation between the two reduplicated
forms (as expected given their semantic opacity) and the most natural, spontaneously produced form was always the
internal reduplication of the rightmost member. One possibility is that those which accept leftmost edge
reduplication have a V-V compound as their historical source (rather than an N-V compound), although
etymological information is not available which would allow us to confirm this hypothesis.
20
j. iva’a-chaka ‘embrace, hugging’ iva’a-cha-cha’e < iva’a ‘hug’ ?
k. iva’a-nama ‘cradling, embracing’ iva’a-na-nama < iva’a ‘hug’ ?
l. kuhtiachi ‘Hateful,mean, kuh-ti-tiachi
awful person or animal’
m. kutsaite ‘dusk, early evening’ kut-sai-saite < kut- ‘dark’?
n. machu’unama ‘hold in hands, grasp’ machu’u-na-nama
o. maukaroa ‘rise in the early mau-ka-karoa
dawn or before dawn’
p. naamuke ‘drunk, dizzy’ naa-mu-muke < naamu ‘cloud‘ ?
q. tekipanoa ‘work’ teki-pa-panoa < tekil- ‘job’?
Reduplication in these pseudo-compounds appears to target an internal head, but our
native speaker consultants, queried about the possible sources of these expressions, indicated that
they were fixed forms with no internal analysis. When further asked about the potential partial
etymologies noted below (which we proposed based on phonological relatedness and semantic
plausibility) they explicitly denied the relatedness of the forms.
It is important to note that while most of the words in (26) are composed of three or four
syllables, this kind of internal reduplication is not required for all words above two syllables; i.e.
this internal reduplication is not strictly phonological. Verbs ending in the derivational suffixes
illustrated in (25) above frequently reach lengths of 4-5 syllables but nonetheless reduplicate on
the left edge; some other trisyllabic forms with initial reduplication are illustrated in (27):
(27) a. Itepo aman si hu-hu’uwaasu
3.PL there very RED-freeze
‘Over there we really freeze!’
b. A’apo siime taewata hi-hiwihte
3.SG always day RED-saw
‘S/he saws (wood) all day long.’
21
c. Uu uusi muni-m chive-chivehta
The child bean-PL RED-spread.out
‘The child is spreading out the beans.’
As shown in (27), the initial reduplicant can vary between one syllable and two syllables, which
for the most part is a stem-specific lexical choice.
With the above considerations in mind, we conclude that the internal reduplication in the
examples in (26) indicates some kind of complex morphological structure, where the
reduplication process apparently targets the head of an opaque compound. As in the NI cases
from the beginning of section 3, the edge of the phonological word/derived stem is not itself the
domain for the attachment of the reduplicative morpheme.
3.4 Interim conclusion: Head-marking
Above we have seen the interaction between reduplication and a variety of compounding and
derivational processes creating complex verb forms in Hiaki. In each case, we have seen that
reduplication targets the rightmost member of the complex verb. We propose that this internal
inflection is a type of head-marking, in which the reduplicative affix applies to the head of a
complex morphological form, even within a single phonological word. We now turn to consider
the theoretical implications of this phenomenon, as well as discussions of similar data in other
unrelated languages.
4. Implications for standard theoretical architectures
The empirical evidence surveyed above represents a case in which lexical integrity appears to be
violated. The Hiaki compound and derived forms we have surveyed are indubitably single
phonological words, exhibiting many characteristics which both language-internally and cross-
22
linguistically are canonical hallmarks of wordhood.5 Nonetheless, a regular inflectional process
appears to target a subconstituent within these derived forms – a process which would normally
target the left edge of a verb, but which in this case targets the left edge of the rightmost element
in a derived verb. This results in an inflectional affix which intervenes between the components
of the derived form.
This situation is not unique to Hiaki, of course. Similar facts are observed for inflectional
reduplication in four languages that we know of. In two other Uto-Aztecan languages, Hopi and
Classical Nahuatl, examples analogous to the Hiaki cases are attested, as illustrated in (28).
Example (28)a shows reduplication inside a nominalized N-V compound from Classical Nahuatl;
(28)b shows reduplication on the head of a V-V compound from Hopi:
(28) a. ixmjmjqujnj
Ø-ix-mi-miqui-ni
3.SG.SUBJ-eye-RED-die-AG.NOM
‘It is one which is blinded (by strong light)’ (lit. ‘It is one whose eyes die’),
(re: the gopher/toçan)
(Classical Nahuatl, Florentine Codex, Book 11: 16)
b. lavay-ho-honaq-lawu
talk-RED-be.erratic-CONT
‘He was just jabbering’ (Hopi, Hopi Dictionary 1998:202)
Both of these language also allow internal reduplication of certain suffixal heads of complex
forms:
(29) a. Auh in jtlaqual mjchtepitzitzin mjchcocone
auh in i-tlaqual mich-tepi-tzi-tzin mich-co-cone
and DET its-food fish-offspring-RED-REV fish-RED-child
‘(and) its food is small fish, baby fish’ (re: the tlacamichin, a type of fish)
5 There is a large literature on the issue of “wordhood”. With respect to the notion 'phonological
word', see the review in Hall 1999. For relevant discussion about the relationship of the
phonological word to the lexeme and/or syntactic terminal ‘word’,, see among many others such
works as Haspelmath 2011, Aronoff 1976, Lieber 1980, Williams 1981, Selkirk 1982, Farmer
1980, Lapointe 1980, 1981, Newmeyer 1986, DiSciullo and Williams 1987, Carstairs-McCarthy
1992, and of course Marantz 1997.
23
(Classical Nahuatl, Florentine Codex, Book 11: 58)
b. ööqa-to-to-yna
bone-RED-CAUS-CAUS
‘She fastens a new stem [in a basket]’ (Hopi, Hopi Dictionary 1998: 362)
Outside Uto-Aztecan, similar instances of reduplication appearing inside derived forms
are analyzed for Sanskrit by Stump (2001); similar cases are discussed from Bahasa Indonesian
by Sato (2010). A Sanskrit example is presented in (30):
(30) pary-a-da-dhat < pari-dha- ‘put around’
around-PRET.AUG-RED-put
‘S/he was putting (it) around (something)’ (Stump 2001: 110; our translation)
These internal reduplication cases are a sub-case of the larger phenomenon of internal
inflectional head-marking of derivationally complex forms, which, although uncommon, are far
from unattested cross-linguistically; see the extensive documentation in Stump (2001: 96-137),
as well as related discussion by Harris (2000, 2002), Sato (2010), and others. Here, we will
specifically consider the theoretical implications of the Hiaki data described above, but many of
the issues raised will of course be relevant to the analysis of head-marking generally; we do not,
however, propose to consider the whole range of head-marking facts here.
First we explore the possibilities within established standard frameworks, both Lexicalist
and syntacticocentric, and conclude that neither family of approaches can implement head-
marking without supplementation by additional mechanisms.6 We then go on to consider what
such supplementation might consist of for each approach, reviewing the proposal of Stump
6 We will mainly discuss theories for which we can find specific proposals dealing with the relationship of
inflection, derivation and compounding in the architecture. Lieber and Scalise (2006) propose a revised Lexicalist
architecture to allow the morphology limited access to the output of phrasal syntax, but we are unclear on whether
their architecture implements any strict restrictions on the interaction of derivational and inflectional processes, so
we do not discuss it here. Insofar as Lieber and Scalise’s approach is strongly Lexicalist, as DiSciullo and Williams
(1987), the remarks about the latter may be relevant to the former as well.
24
(2001) within a word-based approach, and exploiting the mechanisms available in Distributed
Morphology to propose an account within a syntacticocentric framework.
We will assume below, without argument at this point, that reduplication involves the
affixation of a morpheme (“RED”) that triggers some kind of copying process on a stem. In this
we follow many others working within a variety of different theoretical frameworks: e.g.
Moravcsik (1978); Marantz (1982); McCarthy and Prince (1986, 1993, 1995); etc. Other views
are possible, e.g. reduplication occurring as a phonological rule (e.g. Aronoff 1976, Raimy 2000,
Frampton 2009). For a defense of the piece-based view of reduplication over process-based
views, see Haugen (2008), among others. We think that our ultimate conclusions do not ride on
this particular assumption; see our further discussion in section 6 below for what is at ultimately
at stake, and how theories such as Raimy (2000) and Frampton (2009) could be compatible with
the analyses that we present in the sections to follow.
4.1 Weak Lexicalist Architectures
In Lexicalist theories (e.g. Sapir 1911; Mithun 1984, 1986; DiSciullo and Williams 1987; Rosen
1989; Mithun & Corbett 1999), NI is the morphological (“lexical” or pre-syntactic) process of
compounding a nominal root onto a verbal stem. Under such a view, this morphological process
creates a noun-verb compound that names some culturally-significant concept and which is
stored as its own listeme in the lexicon (see e.g., Mithun and Corbett 1999: 68).
Given a division of labor between morphology and syntax like that of Aronoff (1976) or
Anderson (1982), where derivation is lexical and inflection is syntactic (the “Weak Lexicalist
Hypothesis”), the conception of NI as a pre-syntactic lexical process makes a strong prediction
about how NI should interact with an inflectional process of reduplication: this inflectional
process, like others, should not be able to operate inside this compound. The Weak Lexicalist
25
Hypothesis derives peripherality of inflectional affixes as a natural consequence, and thus
predicts that reduplication will apply to the edge of the compound. Under the Weak Lexicalist
Hypothesis, reduplication of NI structures should only operate on the edges of a compound stem
(i.e. on the left-edge for prefixes, or on the right-edge for suffixes), rather than targeting some
sub-constituent of the N-V compound (e.g. the noun or the verb only). This theory thus
incorrectly predicts that Hiaki NI will yield *ma-maso-peute ‘*RED-deer-butcher’ rather than
the actual attested form, maso-peu-peute ‘deer-RED-butcher’.
One language that potentially illustrates this expected interaction of reduplication with NI
is Paiwan, an Austronesian language spoken in Taiwan. Chang and Wu (2006) argue that
Paiwan has two types of NI, lexical and syntactic, which are identified by two independent
diagnostics. Incorporated nominals undergoing lexical NI do not have case-markers, and such
noun-verb compounds can undergo reduplication (31). In syntactic NI, on the other hand, case-
markers are incorporated with the incorporated nominal, and the noun-verb compound cannot
undergo reduplication(32).7 Both of these diagnostics make the first type look like a lexical
process: case-inflection cannot occur inside the N-V compound, cf. (31)b, and the process of
reduplication appears to target the edge of the compound, cf. (31)b and (33):
(31) a. s<em>a-’uma=aken b. s<em>a-’uma-’uma=aken
go.to-<AF>-home=1.SG.NOM go.to<AF>-home-RED=1.SG.NOM
‘I went home’ ‘I am going home’
(32) a. s<em>a-tjaj-palang=aken b. * s<em>a-tjaj-palang-*palang =aken
go.to<AF>-OBL-Palang=1.SG.NOM go.to<AF>-OBL-Palang-*RED=1.SG.NOM
‘I went to Palang’s place’ ‘I am going to Palang’s place’
(33) s<em>a-pana-pana=aken
—— go.to<AF>-river-RED=1.SG.NOM
‘I am going to the riverbank’
7 An unusual trait of NI in Paiwan is that it only occurs with spatial verbs (e.g. verbs of motion or location).
Reduplication that occurs with such verbs indicates progressive aspect, and appears as a suffix.
26
An important observation about the data provided by Chung and Wu is that the
reduplicant seems to always be co-extensive with the nominal root, which in turn seems to
always be disyllabic. Chung and Wu do not present cases of monosyllabic nominal roots, with
which we might expect to see the reduplication of the monosyllabic nominal and the last syllable
of the verb (assuming that the progressive reduplicant in Paiwan is consistently a disyllabic foot).
For example, with a hypothetical nominal stem ba in the present progressive motion
constructions above, we would expect a reduplicated form something along the lines of (34),
where underlining indicates the hypothetical nominal stem, and where brackets indicate the
portion of the compound verbal stem that is copied by the reduplicant (in bold):
(34) Hypothetical Paiwan Monosyllabic Nominal Reduplication
s<e[m>a-ba]-maba=aken
‘I am going to <whatever is denoted by √BA>)
Forms like the hypothetical example in (34) would provide strong evidence in favor of the view
that the N-V compound behaves as a unified stem that gets inflected after syntax, therefore
supporting the idea that it was created prior to syntax, in the lexicon. In the absence of such data,
however, the possibility that reduplication in Paiwan is only targeting the nominal root cannot
yet be ruled out (though its progressive meaning implies that it must at least semantically apply
to an eventive verbal concept, not a stative nominal one).8 In any case, the interaction of
reduplication with NI in Paiwan looks like a reasonable candidate for a Weak Lexicalist analysis.
On the other hand, the interaction of reduplication and NI in Hiaki, as noted above, is a
problem for a theory which predicts edge-marking as a consequence of its architecture. In order
8 Of course, just because the stem to which the RED affix attaches morphosyntactically is verbal, it need not be the
case that the entire stem with its internal morphological complexity is therefore defined as the Base for the
morphophonological process of reduplication which the RED affix triggers. We can distinguish between the
“Target”, i.e. the stem to which the RED affix attaches, and the “Base”, i.e. that potential sub-set portion of the stem
which is delimited as the morphophonological sub-domain marked as available for copying (for further discussion,
see Haugen 2008).
27
for this critique to have teeth, it is particularly important to establish that the Hiaki derived forms
which exhibit head-marking do count as ‘lexical’ in the relevant sense. We therefore pause here
to comment on two established criteria for distinguishing lexical processes from syntactic ones:
(non-)productivity (4.1.1) and non-compositionality (4.1.2).
4.1.1 Productivity
Smirniotopoulos and Joseph (1998) “take the absence of (a high degree of) productivity as a
clear indicator of a lexical rule” (p. 452, emphasis in original). On these grounds, Hiaki NI
should clearly be considered to be a lexical process, since N-V compounding in that language is
relatively restricted (as noted by, e.g., Jelinek, p.c.), being not nearly as productive as in some
other Uto-Aztecan languages, e.g. Nahuatl (Merlan 1976) or Hopi (Hill 2003).
In contrast, the reduplication patterns that we see in Hiaki compound verbs conform to
the otherwise regular and productive morpho-phonological patterns of prefixal reduplication in
the language, in terms of both form and function. In terms of function, reduplication in Hiaki
(in compound verbs and otherwise) typically has a habitual aspectual meaning, which is usually
characterized as verbal inflection, cross-linguistically as well as in Hiaki itself. Like other
inflectional processes, reduplication is fully productive, and it can be applied to any verb that is
compatible with the resulting meaning. In terms of form, reduplication in Hiaki usually involves
syllabic reduplication (Haugen 2003, 2008, Harley and Leyva 2009), e.g. kupi-tom-tom.te (15)f
vs. lioh-bwa-bwa.ni.a (15)g.9 In addition, we also see distinct patterns of reduplicative
allomorphy, e.g. morphological gemination (cf. 17). This kind of allomorphy is also attested
with other verb forms, and it appears to be the case that which reduplicative morpheme (or
dupleme in the terminology of Spaelti 1997) goes with which verb stem is unpredictable and
9 Just as is also the case with other instances of verbal reduplication in Hiaki, we never see copy into the second
syllable to create a coda for syllabic reduplication; e.g. *li.oh.-bwan-bwa.nia.
28
often must be lexically listed (Haugen 2003, 2008, Harley & Leyva 2009); this is similar to other
irregular inflectional processes cross-linguistically.
By the productivity criterion, then, Hiaki NI, which is not productive, is clearly lexical,
and reduplication, which is productive, is clearly inflectional.
4.1.2 Compositionality
Similarly, Smirniotopoulos and Joseph write, “The output of a syntactic rule should show
compositional semantics, so that the meaning of the whole is composed from the meaning of its
parts. By contrast, the output of a lexical rule can be noncompositional in its semantics and thus
can show meanings that differ in ways that are unpredictable in relation to the meanings of the
individual parts composing it” (p. 452). On this criterion, too, these Hiaki compound forms
which exhibit head-marking, are clearly noncompositional; as we have shown above, many of
them have idiomatic meanings—see, for example, (15)d, e, and f above —and some are even
composed entirely of cran-morphs, whose sub-parts do not contribute any detectable independent
meaning to the meaning of the whole, as illustrated in (26) above.10
This point is also made by
Dedrick and Casad (1999:161) in their discussion of lexicalization and non-predicability of
meaning in N-V compounds.
In contrast, the semantic contribution of reduplication to these compound forms is
entirely consistent with the contribution of reduplication to verb forms elsewhere in the
language. As emphasized in section 2.1, the pluractional semantics of reduplication on verbs is
identical in compound and non-compound forms, with three primary interpretations, the most
10
Anticipating some discussion below, we would like to point out here that much recent work within Distributed
Morphology (DM) does not take the notion of (non)compositionality (nor that of (non)productivity) to be indicative
of syntactic vs. lexical processes (see e.g. Marantz 1997).
29
common being that of habitual aspect. There is no significant idiomaticity involved in
interpreting productively reduplicated verbs in the language.
Hiaki noun incorporation, then, is clearly lexical by the compositionality criterion as
well, while reduplication is just as clearly inflectional.
4.1.3 Interim conclusion for weakly lexicalist frameworks
In sum, theories involving a strict division between lexical derivational processes and syntactic
inflectional processes would strongly predict edge-marking behavior, in contrast to the facts
described above. We conclude that without supplemental mechanisms such frameworks are not
equipped to account for the interaction of reduplication and noun incorporation in Hiaki.
4.2 Syntacticocentric architectures: Baker (1988) on Noun Incorporation
We now turn to consider the predictions concerning the interaction of incorporation and
reduplication made in a basic syntactic analysis of incorporation, like that initially proposed by
Baker (1988). Baker argued that noun incorporation was simply syntactic head-movement, by
which the internal object nominal head-adjoins to the verb that is its sister, and is then carried
with the verb through the syntactic tree in any further head-movements in which the verb may
participate. In such an approach, head-movement is assumed to create a syntactically complex
head which corresponds to a single phonological word at Spell-Out. What would such a theory
predict concerning the interaction of reduplication and noun incorporation?
If the incorporated element, when nominal, originates in the object position (i.e. sister to
the verb, as proposed by Baker), it will be closer to the verb, structurally, than any higher
functional morpheme, such as we assume RED to be, since it is an aspectual morpheme denoting
a plurality of eventualities. Thus, RED likely heads an Aspect projection in the inflectional
30
complex. If the usual incorporation via head-to-head movement is the only mechanism assumed
(al a Baker 1988), it will be impossible to situate RED linearly within the N-V compound which
adjoins to it. The problem for a entirely syntactic analysis using only head-to-head movement is
illustrated in (35) below, using our original example of reduplication with NI in (13) above:
(35) a. TP
DP T′
AspP T°
VP Asp°
NP V V° Asp°+hab
N° V°
Aapo t t maso- peute -RED Ø
He deer- butcher-RED Ø
Here, the habitual Asp° head, which is spelled out by RED, forms part of the complex verbal
head resulting from syntactic head-to-head movement. The sister to Asp° is the complex N-V
constituent formed by incorporation of the N° into the V°. If RED is morphophonologically
specified to be a prefix, and hence is ‘flipped’ to appear to the left of its sister at linearization (or
alternatively simply triggers right-adjunction of the N°-V° complex to Asp°), it should be
prefixed to that complete N-V constituent. On that analysis, barring further assumptions, RED
would not be able to intervene between the verb and its incorporated object under the head-
movement analysis. That is, just as is the case in the weakly lexicalist account, the basic
syntacticocentric approach predicts edge-marking inflection for Hiaki inflectional reduplication.
This is clearly inconsistent with the facts shown above, and we conclude again that such a
framework, absent supplemental mechanisms, is also unable to account for the interaction of
noun incorporation and reduplication in Hiaki.
31
4.3 Strong Lexicalist Architectures: DiSciullo & Williams (1987)
Interestingly, the strongly Lexicalist framework articulated by DiSciullo and Williams (1987)
provides the tools to accommodate head-marking phenomena within compounding more easily
than the frameworks described above. In contrast to the Weak Lexicalist architecture described
in section 4.1, in the strong lexicalist architecture all word-formation operations, both
inflectional and derivational, are contained within a single module of the grammar. This allows
the framework to more easily accommodate apparent interleaving of these processes; indeed, the
apparent inflection/derivation distinction is argued to be epiphenomenal within the framework.
DiSciullo and Williams (1987: 25) write, “The real generalization about inflectional
affixes is that they must appear in head position, not that they must appear ‘outside’ all other
word formation—the latter is partly a consequence of the former, although there are cases in
which the former holds but the latter does not.” The Hiaki data would appear to be one such case.
Let us consider how it might be treated in this framework.
In the Strong Lexicalist architecture proposed by DiSciullo and Williams (1987),
compounding involves the creation of a morphological unit that derives its features from the
percolation of the features of the head. For example, consider the underlying structures for the
meaning contrast inherent to the English compounds parts-supplier vs. part-suppliers, shown in
(36) (based on DiSciullo and Williams 1987: 24-5):
(36) a. N[sg] b. N[pl] 2 2 N[pl] N[sg] N[sg] N[pl]
parts supplier part suppliers
[partspl-suppliersg]sg vs. [partsg-supplierspl]pl
Fully inflected nominals, already bearing number features, are the input to the compounding
process; this is possible because of the non-segregated nature of the lexical module in this
32
architecture. Verbal inflection will agree with the number of the entire compound, based on the
number inherited from the head of the compound. In no case should a verb be able to “see into”
the compound to note that there is a second nominal with a potentially conflicting number
specification; however, such inflection-within-derivation is perfectly legitimate in the
framework. For the purposes of the syntax, the feature specifications for the non-head of the
compound are completely irrelevant.
We can see how such an analysis could approach the case of reduplication in N-V
compounding in Hiaki. The rightmost component of the compound, the verb, would enter the
compound already inflected for habitual aspect, i.e. reduplicated; the non-head would then
compound with the inflected verbal head to produce the head-marked compound V°. The
aspectual features of the head would then determine the aspectual properties of the whole by
percolation, appropriately.
(37) a. Step 1: Inflection for habitual aspect:
V[+Hab]
Asp[+Hab] V°
RED- sua
(su-) kill.PL.OBJ
b. Step 2: N-V compounding:
V[+Hab]
N V[+Hab]
kuchu susua
fish RED-kill.PL.OBJ
Despite the ability of the framework to generate the relevant form, however, we see at
least two significant problems with the approach. First, the framework does nothing to prevent
the non-head element in a compound from exhibiting inflectional affixation, thereby permitting
33
the generation of examples like parts-supplier or parks commissioner.11
Such affixation cannot
affect the featural properties of the whole compounded word, of course, since it is on the non-
head element, but no mechanism for outlawing such affixation is provided or, presumably,
desired. This leaves us without an account for the strictly uninflected character of the left-hand
member of the Hiaki N-V compounds: incorporated forms like *kuchu-m-sua ‘fish-PL.OBJ-kill’
are always ungrammatical.
The second question that the Hiaki data raise for the DiSciullo and Williams approach
has to do with the fact that reduplication is prefixal in Hiaki, and yet it behaves as a head,
contributing its features to the complex form in which it is contained. This contravenes the
Right-Hand Head Rule of both Williams (1981) and DiSciullo and Williams (1987: 26, 81). We
take the Right-Hand Head hypothesis to be counter-exemplified by the Hiaki data as well as
numerous other cases from languages around the world, but we will not consider the implications
of this further here.
4.3.1 Interim Conclusion for Standard Architectures
In this section we have argued that two standard theoretical architectures, the Weak Lexicalist
approach and the syntactic approach of Baker (1988), are unable to appropriately place Hiaki
aspectual reduplication inside of the word formed by NI and other compounding processes.
These architectures would need some kind of supplemental machinery to account for these
11
DiSciullo and Williams (1987) do not, apparently, regard examples such as *choirs-boy or *rats-eater to be
ungrammatical, as their framework allows the free generation of such forms. However, a significant literature on
when and why inflectional marking is (im)possible within English compounds, beginning with the level-ordering
work of Kiparsky (1982), has been concerned with exactly how to rule out such cases, which seem flatly
ungrammatical to most English speakers. It is perhaps worth noting that the Kiparsky level-ordering framework
faces the same architectural issues as the Weak Lexicalist frameworks do with respect to the Hiaki data, since it
posits a strict ordering between earlier level processes (such as compounding) and later inflectional processes (such
as Hiaki aspectual reduplication).
34
phenomena. In the next section we will discuss some architectural accommodations that have
been proposed which can better model the Hiaki data.
The Strong Lexicalist architecture designed by DiSciullo and Williams, on the other
hand, is able to account for these data since it does not posit a strict separation of derivational
and inflectional processes. However, the Hiaki data raise other problematic issues for the Strong
Lexicalist Hypothesis: namely, that inflection is strictly forbidden on the non-head member of
compounds in Hiaki, and that reduplication can form a Left-Hand head. Neither of these facts
seem to accord with the predictions made by the Strong Lexicalist theory proposed by DiSciullo
and Williams (1987).
5. Architectural Accommodations
Both weakly-lexicalist word-based models and syntax-only models, then, face problems in
coping with head-marking reduplication inside N-V and V-V compounds in Hiaki. Below, we
review a proposal within the word-based Paradigm-Function Morphology (PFM) framework by
Stump (2001) to accommodate similar cases in Sanskrit (§5.1); he introduces a distinction
between Root-Root compounding and Word-Word compounding, which enables him to account
for both edge-marking and head-marking inflectional behaviors. We then turn to consider what
type of supplementation is needed in a syntax-based model to account for the Hiaki data above
(§5.2), proposing that the operation of Local Dislocation (Embick and Noyer (2007)’s updated
implementation of Marantz 1984’s Morphological Merger) can provide the necessary tools to
capture the patterns observed.
35
5.1 Word-based approaches: Stump (2001) and PFM
The most developed word-based approach to inflectional morphology is the general family of
Word and Paradigm models. Several alternative versions of such models have been proposed,
including prominent proposals by Mathews (1972) and Anderson (1992). We will consider here
the more recent implementation developed by Stump (2001), Paradigm Function Morphology
(PFM), which has an advantage over previous WP theories in that it provides a more satisfactory
account for the morphological inflection of heads than earlier theories.
In PFM, word-forms are derived through the application of rules in a paradigm function
to a lexeme, which generates inflected forms for the lexeme corresponding to each cell in a
language’s paradigm space. Stump introduces a three-way distinction between the types of rules
which produce derived lexemes. The head-marking behavior (or lack thereof) in the inflected
forms of a paradigm function applied to a given derived lexeme depends on the particular type of
rule which produced the lexeme in the first place. “Word-to-word” rules derive lexemes which
exhibit head-marking; “Root-to-Root” rules derive lexemes which exhibit edge-marking. A third
type, “Word-to-Stem” rules, produce lexemes which exhibit double marking (i.e. the inflectional
rules apply to both the head and the edge of the derived lexeme). 12
12
These two patterns suggest a potentially larger typology of inflectional patterns: if edges and heads can each
individually be marked, and they can also be marked simultaneously, one might in addition expect to see cases
where either head, edge, or both could be marked for inflection. Such a case may be illustrated by the Pima
reduplication data presented by Munro and Riggle (2004). In Pima nominal compounds marked for plural,
reduplication can apparently target any member of the compound or any subset of members of the compound, up to
and including each member of the compound. A 5-stem nominal compound like that illustrated in (i) below has, in
theory, 31 possible plural forms. The form illustrated below shows the case in which all stems receive the
inflectional marking:
(i) [lil-mìmida] -hoahas-hàha’a] -[dádagkuanakud:] (Pima, Munro & Riggle 2004 [18])
[glass] -[baskety-jar] -[wiper]
‘glass dish cloth’
Interestingly, there are no scope effects for plural marking in Pima. Rather, tthere is free variation: reduplication of
any or all of the stems in the compound makes the entire compound plural. We will not discuss this pattern further
here except to note that it is not clear to us how the PFM account would extend to these facts.
36
An example of a Root-to-Root derivative given in Stewart and Stump (2007: 407) are the
Breton forms in -ad ‘-ful’; when a form like ti-ad, ‘house-ful’, is inflected for plural, the plural
rule applies to the edge, giving ti-ad-où, ‘house-ful-PL’, rather than to the head, which would
produce *tiez-ad, ‘house.PL-ful’. In contrast, Sanskrit preverb-verb compounding is a Word-to-
Word rule. Hence after a preverb such as vi- ‘away’ is compounded with a verb like gam, ‘go’, to
produce vi-gam, the derivative will exhibit head-marking behavior, so, e.g., the prefixal augment
a- as well as tense/aspect inflection is attached to the head verb –gam, giving vi-a-gacchat ‘s/he
goes away’. The intervention of the inflectional augment prefix inside the complex derivative is
predicted by the fact that it was produced by a Word-to-Word rule.
With respect to inflectional reduplication, PFM thus predicts the pattern seen in example
(30) above, repeated below for convenience:
(38) pary-a-da-dhat < pari-dha- ‘put around’
around-PRET.AUG-RED-put
‘S/he was putting (it) around (something)’ (Stump 2001: 110; our translation)
Here, within a preverb-verb compound, prefixal reduplication applies to the verbal head,
intervening between the preverb and the verb along with the prefixal augment a-. The Hiaki data
are equally amenable to this approach. This analysis would entail that the N-V compounds, the
V-V compounds, and the V-suffix derived forms which exhibit head-internal reduplication are
all produced by Word-to-Word rules; their head-marking behavior with inflectional reduplication
would then be expected.
While the PFM approach can indeed account for the basic patterns described above,
certain of Stump’s claims about the properties of head-marking within his framework may be
called into question when additional data from Hiaki are considered. A central plank in the
theoretical platform Stump espouses is the Paradigm Uniformity Generalization (PUG),
37
according to which “head marking is an all-or-none phenomenon: If a root ever exhibits head
marking in its inflectional paradigm, it always does.” (Stump 2001:109).
For example, the inflectional paradigm function for the Sanskrit verb above specifies
both inflectional reduplication and inflectional prefixation. Due to the PUG, both of those
processes must apply to the verb which is the head of the complex lexeme. The PUG predicts
that a form behaves uniformly with respect to all inflectional rules: one rule cannot head-mark
while another edge-marks. Stump (2001: 133) is helpfully explicit about the type of evidence
which would genuinely disconfirm the PUG. Such potential counterevidence is offered under the
guise of Pseudo-Sanskrit, in which reduplication targets the verb while the augment a- targets the
left edge of the complex form. If the PUG is a valid universal generalization about word-
formation, and specifically the structure of inflectional paradigms, then this kind of disparate
inflectional marking should be impossible. Stump illustrates this hypothetical illicit pattern with
the following form:
(39) “*Pseudo-Sanskrit”
*a-pary-da-dhat < pari-dha- ‘put around’
PRET.AUG-around-RED-put
‘S/he was putting (it) around (something)’
In fact, however, Hiaki compound verbs containing object clitics provide a case
essentially identical to the Pseudo-Sanskrit counterexample above. In Hiaki, object clitics are
inflectional elements which must appear prefixed to the main verb of the clause in which they
occur, as illustrated in (40)a. They may not be separated from the main verb by any nonverbal
material (40)b, even particles which are important to the entire predicate’s meaning and which
otherwise must appear adjacent to the main verb themselves (40)c (see discussion in Dedrick and
Casad 1999: 269).
(40) a. Vempo aman aa=vicha-k
38
3.PL.NOM there 3.SG.OBJ =see-PERF
‘They saw him/her over there.’
b. *Vempo aa=aman vicha-k
3.PL.NOM 3.SG.OBJ =there see-PERF
c. nat am=totta-ka
on.top 3.PL.OBJ=pile-PPL
‘piling them on top of one another’ (Dedrick and Casad 1999: 271)
The object clitics, then, are inflectional elements which prefix to their verb, just like
reduplication does; when attached to a non-compound verb form, both appear as prefixes to the
verb stem, as expected:
(41) Kat=ee unna kusisi aa=’e-’eta
NEG.IMP=2.SG.NOM too loudly 3.SG.OBJ =RED-close
‘Don’t close it too loudly!’
However, when reduplication and an object clitic are attached simultaneously to a compound
verb, they behave exactly like Pseudo-Sanskrit: i.e. the object clitic attaches to the left edge of
the complex form while reduplication targets the head. We saw this pattern illustrated above in
example (16), which we repeat for convenience below as (42). The grammatical form in (42)a is
equivalent to the Pseudo-Sanskrit example above. (42)b would be expected in a uniformly edge-
marking Pseudo-Hiaki which conformed to the PUG; (42)c would be expected in a uniformly
head-marking Pseudo-Hiaki', also conforming to the PUG; both are flatly ungrammatical.
(42) a. Irene am=pan-ho-hoo-ria
Irene 3.PL-bread-RED-make-APPL
‘Irene is always making bread for them.’
b. *Irene am=pa(n)-pan-hoo-ria
Irene 3.PL- RED-bread-make-APPL
c. *Irene pan-am=ho-hoo-ria.
Irene bread-3.PL=RED-make-APPL
39
So, while the head-marking mechanism proposed by Stump can accommodate the basic
Hiaki cases, assuming that the verb compounds are produced by a Word-to-Word rule,13
the way
the system interacts with paradigm functions would have to be relaxed in order to allow for some
inflectional processes to be specified as edge-marking with even Word-to-Word derivatives
while others are specified as head-marking.
We conclude that the PUG cannot be maintained in its current form in the face of
examples like (42). This represents a serious challenge to the PFM architecture, which conspires
to derive the PUG as a sub-case of the Head Application Principle (HAP), a supposed universal
of morphological structure. As Stump himself indicates, one can conceive of evidence which
genuinely disconfirms the PUG, hence also the HAP. We submit that the Hiaki pattern of object
cliticization and reduplication constitute such evidence. We therefore suggest that the PUG
cannot be a universally valid generalization about the structure of inflectional paradigms.
We now turn to consider a potential account of this data in a syntacticocentric framework
supplemented with certain purely morphological operations, Distributed Morphology (Halle and
Marantz 1993 et seq.).
5.2 Syntacticocentric approaches: Distributed Morphology
Above, we sketched the syntactic account of noun incorporation proposed by Baker (1988), and
showed that, like the Weak Lexicalist account, it too produces the incorrect prediction that
13
Stump claims that both the root and the resulting derivative of a Word-to-Word Rule must be either both roots or
both nonradical words (Stump 2001:117). Nonradical words are forms extracted from the fully-inflected paradigm
of a lexeme. In the case of most Hiaki examples, we assume that within PFM, the root and the derivative would
themselves both have to be roots. However, in example (24)c above, we exhibit a case where the root, as well as the
derivative, exhibits reduplication as an inflectional marker of habitual aspect. In Stump’s terminology, one can
conclude that this case would have to involve two nonradical words; the inflected root form would be drawn from
the output of the paradigm function applied to the basic root, which produces nonradical words. However, the
leftmost member of this complex form is not a nonradical word itself; rather, it is still a bound form: no-nok-; the
corresponding habitually inflected free form is no-noka, ‘RED-speak’. This may prove puzzling for the definition of
‘Word-to-Word Rule’ in PFM.
40
reduplication in Hiaki should be edge-marking, rather than head-marking, since the incorporated
N-V form is adjoined as a whole to the Asp° head realized by the RED prefix. The RED prefix
should then attach to the entire N-V subconstituent, not just to the V. Similar remarks apply to
the V-V compound and V-suffix cases in their interaction with reduplication. A Baker-style
analysis of a V-V compound like that in (24)a above is illustrated below; the embedded verb
raises to adjoin to the matrix verb, and again both verbs move to Asp°, predicting edge-marking
behavior:
(43) AspP
VP1 Asp°
DP V1' RED
HAB
Nee VP2 V1°
1sg.NOM
DP V' V2° V1°
aa tV2 nok- ii'aa
3sg.ACC speak want
Nee aa=nok-i’-ii’aa
1sg.NOM 3sg.ACC=speak-RED-want
‘I (habitually) want him to speak.’
We will propose that the interaction of reduplication and verb-compounding in Hiaki is best
analyzed, within the Distributed Morphology framework (Halle and Marantz 1993, 1994; Harley
and Noyer 1999; Embick and Noyer 2007, a.o.), as an example of the post-syntactic operation
Local Dislocation, of the type discussed in Embick and Noyer (2007).
Following head-to-head movement of the complex verb to Asp°, the Asp° head will have
the internal structure illustrated below:
41
(44) Asp°
V1° Asp°
V2° V1°
nok- -ii’aa -RED
speak want HAB
At Morphology, this structure undergoes Vocabulary Insertion and Linearization. The
insertion of the RED morpheme will trigger a morpheme-specific operation of Local Dislocation,
in the sense of Embick and Noyer (2007). This operation prefixes the RED morpheme to the first
lexical verb to its left. The structural operations within the complex Asp° head are represented in
bracketed notation in (45) below:
(45) a. [[ V° V° ]V° Asp° ]Asp° (Complex Asp° head — Output of syntax)
b. [[ nok [ii’aa-RED] ] Asp° (Insertion of Vocabulary Items, Linearization)
c. [[ nok [RED-ii’aa]Asp° ] Asp° (Local dislocation of RED and ii’aa)14
d. [[ nok [i’-ii’aa]] (Phonological content of RED computed by
copying from the base.)
That is, in order to accommodate the Hiaki facts in a syntacticocentric approach, the syntax must
be supplemented by some operations specific to the morphology component; syntax alone does
not suffice. Under the version of the Y-model of grammar assumed by DM, (45)a is the output of
syntax proper, while (45)b and (45)c occur in the post-syntactic module of Morphological
14
Within DM, this example illustrates an interesting feature of the relationship between the Headedness Parameter
(see, e.g. Baker 2001), which linearizes syntactic terminal nodes, and the prefixal or suffixal status of particular
Vocabulary Items (VIs), which we take to be specified by Alignment constraints operating on those specific VIs.
Linearization must have applied before Local Dislocation so that the heads undergoing Local Dislocation are
linearly adjacent. We conclude that the prefixal or suffixal status of an affix has its effect within a structured,
morphological string that is already linearized, but the notions “prefix” or “suffix” do not drive linearization itself.
That is, Alignment seems to be separate from Headedness, as we would expect: Hiaki is head-final (a general
Linearization constraint), but RED is prefixal.
42
Structure. The operation in (45)c, which, in conjunction with (45)b, resolves the reduplicant
linearization conundrum, is taken to be a PF operation.15, 16
The same sequence of derivation can apply to derive the examples in (24b) and (24c)
above, where reduplication applies to the embedded verb, on the assumption that in those cases,
the complement to –ii'aa is not a VP, but instead includes an entire AspP with a habitual RED
morpheme in its head. That embedded RED morpheme will scope only over the embedded verb,
and will undergo local dislocation to the right of the embedded verb in such a case, deriving the
patterns with low-scope reduplication on the leftmost element of the complex predicate, or (if
there is habitual aspect in both the matrix and the embedded clauses) reduplication on both,
interpreted, as indicated above, in both places.
Below we show the same series of operations as it would apply to our initial example of
reduplication interacting with Noun Incorporation, in (13) above, based on the Baker-style tree in
(35), maso-peu-peute, ‘deer-RED-butchering’:
(46) a. [[ N° V° ]V° Asp° ]Asp° (Complex Asp° head — Output of syntax)
b. [[ maso [peute-RED] ] Asp° (Insertion of Vocabulary Items, Linearization)
c. [[ maso [RED-peute]Asp° ] Asp° (Local dislocation of RED and peute)
d. [[ maso [peu-peute]] (Phonological content of RED computed by
copying from the base.)
15
In fact, different DM proposals distinguish several different operations similar to Local Dislocation, including
"Merger Under Adjacency" (Bobaljik 1994), which itself is similar to Mithun (1984)’s Compounding by
Juxtaposition. The proposal in the text represents just one of several possible DM analyses of these constructions,
each with its own set of consequences for the analysis of hyponymous objects, object clitics, etc. We will not
consider alternatives here, but the proposal in the text represents the optimal account given the larger empirical
picture, according to our best current understanding. See Harley (2010) for a review of some of the range of
analytical possibilities made available by the interaction of these operations with syntactic head-movement. 16
In the end, then, our account does not ascribe ‘head-marking’ reduplication to Hiaki; rather, the reduplicant in
Hiaki morphologically selects or aligns itself with the closest available root to its left. As shown by the mochik case,
the root need not in fact be verbal in character. The head of the constituent to which the aspectual meaning is
attached is actually a v°, projecting vP, in all of the above, while the √ that ends up bearing the inflection is itself
further embedded.
43
The tree illustrating the syntactic structure of the above example in (35) represents a
much earlier view of phrase structure than is commonly assumed in syntacticocentric analyses
today. Most importantly from our current point of view, the representation of peute as heading a
simplex V node is inaccurate, given the considerable developments in the theory of argument
structure syntax in the decades since Baker’s original proposal. Agentive verbs are commonly
now understood to consist of a minimum of two projections: a lexical root which selects the
internal argument (corresponding to the original ‘V’), and an external-argument introducing
‘light verb’ functional projection, which we will notate here as v°. Indeed, the verb pair peu-
te/peu-ta ‘butcher-INTR/butcher-TR’ contains an overt morpheme which itself is plausibly
analyzed as an instantiation of that v° node (though see Jelinek 1998 and Tubino Blanco 2010
for more detailed discussion). That is, the actual base-generated structure, without any syntactic
head-movement, in this approach, does not look like (35) but rather should minimally include the
functional projections below:
(47) TP
T'
AspP T°
Asp'
vP Asp°
DP v'
√P v°
NP √
N°
Aapo maso peu te -RED Ø
He deer- butcher- INTR RED Ø
44
Following movement of the various heads (and the subject for nominative case), the output of the
syntax at Spell-Out, prior to Morphology, will have (minimally) the following form:
(48) TP
DPi T'
AspP T°
Asp' Asp° T°
vP tAsp v° Asp°
ti v' √ v°
√P tv N √
NP t√
tN
Aapo maso- peu-te -RED-Ø
He deer-butcher-INTR-RED-Ø
Of interest, of course, is the structure under the complex T° head, containing the result of head-
movement through the verbal extended projection. The reduplicative morpheme, RED, which is
inserted to realize the Asp° node, is subject to Local Dislocation. However, this dislocation does
not apply to RED and its immediate neighbor, which is the light verbal morpheme –te. Rather,
RED is dislocated leftward until it is prefixed to the first available lexical root morpheme, √peu-.
In footnote 14 above, we suggested that Local Dislocation could be conceived of as the
application of a Vocabulary-Item-specific Alignment constraint; this would be a perspicuous
implementation of what appears to be a straightforward subcategorization requirement of the
RED morpheme: specifically, it requires a lexical root as its host.
This implementation also predicts that the RED affix will not necessarily be sensitive to
45
the lexical category of the item which hosts it; under standard DM assumptions, it is looking for
an acategorial lexical √ to attach to, not a particular category.17
This prediction appears to be
borne out in another interesting corner of Hiaki grammar, the possession/use construction. In this
construction, a nominal root is inflected with verbal morphology, receiving an interpretation of
‘possessed N’. An example is given in (49) below:
(49) Huan mochik-e-k
Juan turtle-PHAVE-PERF
‘Huan has a turtle/turtles.’ (Lit: ‘Huan is turtled’.)
Reduplication in this construction targets the √ of the nominal mochik ‘turtle’:
(50) Huan mo-mochik-e
Huan RED-turtle-PHAVE
‘Huan usually has turtles.’ (Haugen 2004: 264)
This supports the notion that the RED affix subcategorizes for √ morphemes: when the closest
available √ in the complex head happens to be a nominalized root, rather than a verbalized one,
that is the √ with which RED undergoes Local Dislocation/Alignment.18
Under the present account, we can conclude that the difference between the verbal affixes
which support reduplication such as -ii’aa ‘want’, described above in section 3.2, and those such
as –tua ‘CAUS’ or -ria, ‘APPL’, which do not, is that the former retain their lexical roots,
despite being on a grammaticalization path which has restricted them to bound positions and
which potentially could ultimately result in their reanalysis as heads of functional categories such
17
Because of the syntactic and semantic requirements of the Asp° head which RED is the realization of, this RED
morpheme will necessarily appear in predicative contexts in Hiaki; however, under this treatment, the RED
vocabulary item itself does not select for verbs, but rather subcategorizes for/aligns with √ morphemes. 18
Note that within the DM framework’s assumptions, all √s must occur within the scope of a categorizing
morpheme. In many cases this morpheme is null, as is the case with the nominalizing head that presumably
intervenes between the √ and the ‘PHAVE’ marker here. In other cases, it is overt, as with the intransitive verbalizer –
te in Hiaki siute, ‘tear.intr’ or English –ify as in clarify, stupefy.
46
as v°. The latter affixes, which do not support independent reduplication, we presume to head
functional projections in the synchronic grammar.19
The above analysis represents our current best understanding of the optimal approach to
this complex array of facts within a syntacticocentric analysis. Other avenues of analysis
obviously remain, but we hope to at least have shown that the interaction of reduplication and
head-marking is amenable to treatment given reasonably non-contentious assumptions within
such frameworks. The analytical key which the syntacticocentric approach makes available is
that the internal structure of the complex word-form remains accessible throughout the
derivation, given the single-engine architecture of the framework.
6. Conclusion
It should be clear that important architectural issues ride on the correct theoretical approach to
the range of facts described above, and on similar little-studied facts from languages around the
world. Many questions and issues, however, remain open. Here we comment on some general
and specific implications of and questions raised by the discussion above, and indicate briefly
some future directions that the current line of analysis opens up.
We have considered how several distinct grammatical architectures could in principle
approach the Hiaki facts. A Strong Lexicalist approach like that of DiSciullo and Williams
(1987) seems to offer a set of tools which could allow an account of the interaction of the
derivational and inflectional processes we are considering. The relevant feature of the Strong
Lexicalist architecture is the lack of differentiation between derivational and inflectional
processes, which allow us in principle to interleave the two. Interestingly, a unified approach to
inflection and derivation is a feature which Strong Lexicalism shares with single-engine
19
These affixes most likely originally arose from independent lexical verbs which underwent this same
grammaticalization process.
47
syntacticocentric analyses: i.e. all word-building operations, inflectional and derivational, occur
in the same generative subcomponent of the grammar. Thus, architectures which group all types
of word-formation operations into a single component do not encounter the problems that we
have identified for Weak Lexicalist and word-based architectures.
In the latter frameworks, on the other hand, if derivation is treated as creating word-forms
which are input to an inflectional word-formation process, such as a Paradigm Function, head-
marking phenomena become difficult to account for. Given certain additional assumptions, such
as those concerning the formation of derivatives that Stump (2001) puts forward, such
frameworks can accommodate head-marking phenomena. The specific proposal of Stump,
however, runs aground on the mixed character of inflection in Hiaki. Stump’s PUG requires that
a derived form will consistently head-mark or edge-mark (or doubly-mark) with respect to all
inflectional processes of the language. Inflectional systems in which some inflection is head-
marking and some edge-marking within the same form are outlawed. However, exactly such a
case is presented by the Hiaki NI + reduplication cases in combination with the object
cliticization properties of the language. Object cliticization is edge-marking in the very same
forms in which reduplication is head-marking. Consequently, we conclude that a PFM analysis
of these facts is untenable as things currently stand.
The syntacticocentric approach that we advocate also requires some supplemental
assumptions in order to correctly position the inflectional reduplicant inside the compound verb
form. However, these assumptions that we utilize are not novel ones; rather, they are well-
established mechanisms within the Distributed Morphology framework. The strength of a DM
approach is that, like the Strong Lexicalist approach, derivational and inflectional affixation is
accomplished in the same component—in the case of DM, this is the syntax. Hence, the internal
48
structure of complex derived forms remains visible even following further inflectional affixation.
In contrast to PFM, no requirement that complex forms are restricted to either head-marking or
edge-marking behavior is either expressed or implied. The prefixation of the Hiaki object clitic
to the finite tensed verb form, for example, can coexist simultaneously with the head-marking
behavior of the reduplicative prefix. We assume, in fact, that Hiaki object clitics are positioned
by a syntactic mechanism similar to that which positions Romance object clitics next to the
tensed verb: i.e. syntactic clitic movement to TP and subsequent merger of the clitic with the T°
root node (cf. Matushansky 2006) apparently captures the central facts concerning Hiaki object
clitic distribution.
Unsurprisingly, many questions remain. One issue which we leave for future
investigation is the status of reduplication as a process or an affix. We have assumed an affix-
based approach here, following argumentation in Haugen (2008, 2010, 2011). However, an
approach treating reduplication as a readjustment operation triggered by null affixation like that
proposed in Raimy (2000) or Frampton (2009) could also be entertained. Under this latter view,
the aspectual head involved in triggering reduplication would be realized by a null affix which
would trigger stem readjustment.20
Just as in the account proposed here, the stem readjustment
approach would also have access to the internal structure of the derived verb form, and in fact it
would require this: the null-affixation and stem-adjustment approach would still have to make
reference to the ‘left edge of the closest lexical root’ as the domain for the relevant readjustment.
We leave open here the ultimate question of which of these approaches to reduplication should
be preferred within a syntacticocentric framework.
20
Under Raimy (2000)’s approach, this readjustment operation would involve re-linearization, i.e. altering the
precedence relationships between segments of the stem. For Frampton (2009), the operation involves inserting
‘duplication junctures’ into the timing tier, ultimately resulting in multiple links to the segmental material of the
stem.
49
To conclude, the phenomena considered above graphically illustrate the importance of a
‘big-picture’ view of grammatical systems. Noun incorporation has traditionally (within
generative grammar) been the province of morphosyntacticians, while reduplication has been
investigated primarily from the perspective of morphophonology. The interaction of these two
phenomena in a single language has significant implications that go beyond the narrow concerns
of these two sub-domains, and bear much more broadly on the architecture of grammatical
theory.
Acknowledgments
Our first thanks go, as always, to our wonderful Hiaki consultants, Maria Florez Leyva and
Santos Leyva, who patiently provided and explained much of the key data presented here.
Rosario Amarillas also generously contributed much data and commentary. We would also like
to thank the National Science Foundation, which funded the research reported here, under grant
BCS-0446333. Examples from Classical Nahuatl were collected from the Florentine Codex
(English translation by Anderson and Dibble 1963) during the course of the Classical Nahuatl
Project at Oberlin College, via funding from the OC Office of Sponsored Projects, with the help
of OC Research Assistants Miriam Rothenberg and Anne Thompson. Thanks also to Shannon
Bischoff and Mercedes Tubino Blanco for valuable feedback.
References
Anderson, Arthur J.O. and Charles E. Dibble (eds.). 1963. Florentine Codex: General History of
the Things of New Spain by Fray Bernardino de Sahugún: Book 11: Earthly Things.
Santa Fe, NM: The School of American Research and The University of Utah.
Anderson, Stephen. 1982. Where’s morphology? Linguistic Inquiry 13: 571-612.
——. 1992. A-morphous Morphology. Cambridge: Cambridge University Press (Cambridge
Studies in Linguistics 62).
Aronoff, Mark. 1976. Word Formation in Generative Grammar. Cambridge, MA: MIT Press.
Baker, Mark C. 1988. Incorporation: A Theory of Grammatical Function Changing. Chicago:
University of Chicago Press.
——. 2001. The Atoms of Language: The Mind’s Hidden Rules of Grammar. New
York: Basic Books.
Bobaljik, Jonathan David. 1994. What does adjacency do? In The Morphology-Syntax
Connection, ed. H. Harley and C. Philips, 1-32. Cambridge, MA: MIT Working Papers
in Linguistics 22.
50
Carstairs-McCarthy, Andrew. 1992. Current morphology. London and New York: Routledge
Chang, Henry Y., and Chun-Ming Wu. 2006. Nominal incorporation in Paiwan and its
implications. Paper presented at the Conference on Nominal Incorporation and Its Kind
(NIK), University of Ottawa. February 21, 2006.
Dedrick, John M., and Eugene H. Casad. 1999. Sonora Yaqui Language Structures. Tucson, AZ:
University of Arizona Press.
DiSciullo, Anna-Maria, and Edwin Williams. 1987. On the Definition of Word. Cambridge,
MA: MIT Press.
Downing, Laura J. 2003. Compounding and tonal nontransfer in Bantu languages. Phonology 20:
1-42.
Embick, David, and Rolf Noyer. 2007. Distributed Morphology and the Syntax-Morphology
Interface. In The Oxford Handbook of Linguistic Interfaces, ed. G. Ramchand and C.
Reiss, 289-324. Oxford: Oxford University Press.
Escalante, Fernando. 1990. Voice and argument structure in Yaqui. Ph.D. dissertation,
University of Arizona.
Farmer, Ann. 1982. On the interaction of morphology and syntax. Doctoral dissertation. MIT.
Frampton, John. 2009. Distributed Reduplication. Cambridge, MA: MIT Press. (Linguistic
Inquiry Monographs 52).
Halle, Morris, and Alec Marantz. 1993. Distributed morphology and the pieces of inflection. In
The View from Building 20: Essays in Linguistics in Honor of Sylvain Bromberger, ed.
Kenneth Hale and Samuel Jay Keyser, 111-176. Cambridge, MA: MIT Press
——. 1994. Some key features of Distributed Morphology. In Papers on Phonology and
Morphology, ed. Andrew Carnie and Heidi Harley, 275-288. Cambridge, MA: MIT
Working Papers in Linguistics 21.
Harley, Heidi. 2010. Affixation and the Mirror Principle. In Interfaces in Linguistics, ed. R. Folli
and Christiane Ullbricht, 166-186. Oxford: Oxford University Press.
Harley, Heidi, and Maria Amarillas. 2003. Reduplication multiplication in Yaqui: Meaning X
form. In Studies in Uto-Aztecan, ed. Luis M. Barragan and Jason D. Haugen, 105-139.
Cambridge, MA: MIT Working Papers on Endangered and Less Familiar Languages #5.
Harley, Heidi, and Maria Florez Leyva. 2009. Form and meaning in Hiaki (Yaqui) verbal
reduplication. International Journal of American Linguistics 75: 233-72.
Harley, Heidi, and Rolf Noyer. 1999. State-of-the-Article: Distributed Morphology. Glot
International 4: 3-9.
Harris, Alice C. 2000. Where in the word is the Udi clitic? Language 76: 593-616.
——. 2002. Endoclitics and the Origins of Udi Morphosyntax. Oxford: Oxford University Press.
Haspelmath, Martin. 2011. The indeterminacy of word segmentation and the nature of
morphology and syntax. Folia Linguistica 45,1: 31-80
Haugen, Jason D. 2003. Allomorphy in Yaqui Reduplication. In L. Barragan and J. Haugen
(eds.), Studies in Uto-Aztecan. Cambridge, MA: MIT Working Papers on Endangered
and Less Familiar Languages #5, pp. 75-103.
——. 2004. Denominal Verbs in Yaqui. In Z. Estrada Fernández, A. Fernández Garay, and A.
Álvarez González (eds.), Estudios en lenguas amerindias: Homenaje a Ken L. Hale.
Hermosillo, Son: Editorial Unison, pp. 229-267.
——. 2008. Morphology at the Interfaces: Reduplication and Noun Incorporation in Uto-
Aztecan. Amsterdam/Philadelphia: John Benjamins Publishing Company (Linguistik
Aktuell/Linguistics Today 117).
51
——. 2010. Reduplication in compounding contexts: Morphological doubling vs.
correspondence. To appear in the Proceedings of NELS 39.
(http://ling.auf.net/lingBuzz/001368).
——. 2011. Reduplication in Distributed Morphology. Proceedings of the 4th
Annual Arizona
Linguistics Circle Conference (ALC 4). Coyote Papers vol. 18. Tucson, AZ: University of
Arizona Department of Linguistics.
(http://arizona.openrepository.com/arizona/handle/10150/138590)
Hill, Jane H. 2003. Subject number agreement, grammaticalization, and transitivity in the
Cupeño verb construction. In Formal Approaches to Function in Grammar: In
Honor of Eloise Jelinek, ed. Andrew Carnie, Heidi Harley, and MaryAnn Willie,
207-226. Amsterdam: John Benjamins Publishing Company.
Hill, Jane H., and Ofelia Zepeda. 1992. Derived words in Tohono O’odham. International
Journal of American Linguistics 58.4: 355-404.
——. 1998. Tohono O’odham (Papago) plurals. Anthropological Linguistics 40.1: 1-42.
Hill, Kenneth C. 2003. Hopi denominal and noun-incorporating verbs. In Studies in Uto-
Aztecan, ed. Luis M. Barragan and Jason D. Haugen, 215-244. Cambridge, MA: MIT
Working Papers on Endangered and Less Familiar Languages #5.
Hopi Dictionary Project. 1998. Hopi Dictionary/Hopìikwa Lavàytutuveni: A Hopi-
English Dictionary of the Third Mesa Dialect. Tucson, AZ: University of
Arizona Press.
Inkelas, Sharon and Cheryl Zoll. 2005. Reduplication: Doubling in Morphology. Cambridge:
Cambridge University Press. (Cambridge Studies in Linguistics 106).
Jelinek, Eloise. 1998. Voice and transitivity as functional projections in Yaqui. In The Projection
of Arguments: Lexical and Compositional Factors, ed. Miriam Butt and Wilhelm Geuder,
195- 224. Stanford, CA: CSLI Publications.
Kiparsky, Paul. 1982. Word formation and the lexicon. Proceedings of the 1982 Mid-America
Linguistics Conference, 3-29. Lawrence, KS: University of Kansas.
Lapointe, Steven. 1980.A theory of grammatical agreement. Doctoral dissertation. MIT.
Lapointe, Steven. 1981. General and restricted agreement phenomena. In: Michael Moortgat,
Harry van der Hulst and Teun Hoekstra (eds.) The scope of lexical rules. 125-159.
Dordrecht: Foris.
Lieber, Rochelle. 1980. On the organization of the lexicon. Doctoral dissertation. MIT.
Lieber, Rochelle and Sergio Scalise. 2006. The Lexical Integrity Hypothesis in a new theoretical
universe. Lingue e Linguaggio 5.1: 7-32.
Marantz, Alec. 1982. Re reduplication. Linguistic Inquiry 13: 435-82.
——. 1984. On the Nature of Grammatical Relations. Cambridge, MA: MIT Press.
——. 1997. CAT as a phrasal idiom. Ms, MIT.
Matthews, P. H. 1972. Inflectional Morphology: A Theoretical Study Based on Aspects of Latin
Verb Conjugation. Cambridge: Cambridge University Press.
Matushansky, Ora. 2006. Head movement in linguistic theory. Linguistic Inquiry 37: 69-109.
McCarthy, John J. and Alan S. Prince. 1986. Prosodic morphology. Ms, University of
Massachusetts—Amherst and Rutgers University.
——. 1993. Prosodic Morphology I: Constraint interaction and satisfaction. Ms, University of
Massachusetts—Amherst and Rutgers University.
——. 1995. Faithfulness and reduplicative identity. In University of Massachusetts Occasional
Papers in Linguistics 18: Papers in Optimality Theory, ed. Jill Beckman, Laura Walsh
52
Dickey, and Suzanne Urbanczyk, 249-384. Amherst, MA: Graduate Linguistics Students
Association.
Merlan, Francesca. 1976. Noun incorporation and discourse reference in Modern Nahuatl.
International Journal of American Linguistics 42: 177-191.
Mithun, Marianne. 1984. The evolution of noun incorporation. Language 60: 847-95.
——. 1986. On the nature of noun incorporation. Language 62: 32-38.
Mithun, Marianne, and Greville G. Corbett. 1999. The effect of noun incorporation on
argument structure. In Boundaries of Morphology and Syntax, ed. Lunella Mereu,
49-71. Amsterdam: John Benjamins Publishing Company.
Molina, Felipe S., Herminia Valenzuela, and David Leedom Shaul. 1999. Yoeme/English-
English/Yoeme Standard Dictionary. New York: Hippocrene Books.
Moravcsik, Edith. 1978. Reduplicative constructions. In Universals of Human Language vol. 3:
Word Structure, ed. Joseph H. Greenberg, 297-334. Stanford, CA: Stanford University
Press.
Munro, Pamela, and Jason Riggle. 2004. Productivity and lexicalization in Pima compounds.
In Proceedings of BLS 30. Berkeley, CA: Berkeley Linguistics Society.
Newmeyer, Frederick. 1986. Linguistic theory in America, second edition. New York: Academic
Press
Raimy, Eric. 2000. The Phonology and Morphology of Reduplication. Berlin: Mouton de
Gruyter.
Rosen, Sara Thomas. 1989. Two types of noun incorporation: A lexical analysis.
Language 65: 294-317.
Sadock, Jerrold M. 1991. Autolexical Syntax: A Theory of Parallel Grammatical
Representations. Chicago: University of Chicago Press.
Sapir, Edward. 1911. The problem of noun incorporation in American languages. American
Anthropologist 13: 250-82.
Sato, Yosuke. 2010. Complex phrase structures within morphological words: Evidence from
English and Indonesian. Lingua 120: 379-407.
Selkirk, Lisa. 1981. English compounding and the theory of word structure. In: Michael
Moortgat, Harry van der Hulst and Teun Hoekstra (eds.) The scope of lexical rules. 229-
277. Dordrecht: Foris
Smirniotopoulos, Jane C., and Brian D. Joseph. 1998. Syntax vs. the lexicon: Incorporation and
compounding in Modern Greek. Journal of Linguistics 34: 447-488.
Spaelti, Philip. 1997. Dimensions of Variation in Multi-Pattern Reduplication. Ph.D.
dissertation, University of California—Santa Cruz.
Stump, Gregory T. 2001. Inflectional Morphology: A Theory of Paradigm Structure.
Cambridge: Cambridge University Press. (Cambridge Studies in Linguistics 93).
Stewart, Thomas and Gregory T. Stump. 2007. Paradigm Function Morphology and the
Morphology-Syntax Interface. In The Oxford Handbook of Linguistic Interfaces, ed.
Gillian Ramchand and Charles Reiss (eds.), 383-421. Oxford: Oxford University Press.
Stump, Gregory T. 2001. Inflectional Morphology: A Theory of Paradigm Structure. Cambridge:
Cambridge University Press. (Cambridge Studies in Linguistics 93).
Tubino Blanco, Mercedes. 2010. Contrasting causatives: A Minimalist approach. Ph.D.
dissertation, University of Arizona.
53
Tubino Blanco, Mercedes, and Heidi Harley. 2010. Dos tipos de base verbal en Hiaki (yaqui):
Hacia una clasificación morfológica. Memoria del IX Encuentro Internacional de
Lingüística en el Noroeste. Hermosillo: Editoriales UniSon.
Tubino Blanco, Mercedes, Heidi Harley, and Jason D. Haugen. 2009. The syntax of hybrid
verb/affix lexemes and clause fusion in Hiaki (Yaqui). Rice Working Papers in
Linguistics 1: 79-91.
Williams, Edwin. 1981. On the notions ‘lexically related’ and ‘head of a word’. Linguistic
Inquiry 12: 245-74.
Zoll, Cheryl. 2002. Vowel reduction and reduplication in Klamath. Linguistic Inquiry 33: 520-7.
Recommended